内容来源：https://www.quantamagazine.org/how-animals-build-a-sense-of-direction-20260121/

内容总结：

动物如何构建方向感？自然实验揭示大脑“内置指南针”奥秘

在印度洋一座偏远岛屿上，一项突破性的研究首次在自然环境中揭示了哺乳动物大脑如何构建方向感。科学家通过监测六只埃及果蝠在野外飞行时的脑部活动，证实了其大脑内的“头部方向细胞”如同一个稳定的“内置指南针”，该指南针并非依赖地磁场或天体，而是通过识别环境中的地标进行校准。

这项发表于《科学》杂志的研究，为持续数十年的实验室理论提供了关键的现实验证。研究团队选择坦桑尼亚以东约40公里的莱瑟姆岛作为“天然实验室”，该岛面积约四个足球场大小，既能确保蝙蝠无法一览全貌，又便于科学家追踪。研究人员在蝙蝠大脑植入微型记录设备，在2023至2024年的两个实验季中，成功记录了301次飞行的神经数据。

数据分析显示，蝙蝠在探索新环境初期，其头部方向细胞的放电模式较为粗略。但随着对岛屿熟悉度增加，这些细胞很快稳定下来，始终与特定的精确方向保持一致，不会因蝙蝠在岛上的位置改变而重置。这强有力地支持了“全局指南针”假说，即方向细胞在广阔环境中会保持固定的方向指向，如同真正的指南针。

尤为重要的是，研究发现蝙蝠并非依靠月亮位置或地磁场来定向。科学家推测，它们很可能是通过识别海岸线、实验帐篷等固定地标，在大脑中构建心理地图，并以此校准方向感。

这项研究被誉为“该领域50年内都会被谈论的论文”，它首次在真实、复杂的环境中揭示了哺乳动物导航神经机制的工作方式。后续未发表的初步数据进一步表明，在野外环境中，与导航相关的脑细胞编码的信息远比实验室中丰富。

该成果推动神经科学研究范式向真实世界拓展。目前，科学家正尝试在人类身上寻找类似的机制。尽管尚未直接定位到人类的“头部方向细胞”，但初步证据表明我们可能拥有相似的系统。杜克大学的团队已首次在癫痫患者于医院环境中活动时，记录到与导航相关的脑细胞活动。

专家指出，探索更复杂、自然的环境，将帮助我们发现实验室中无法观测到的大脑奥秘。这项岛屿研究不仅解答了关于方向感的基本科学问题，也为理解人类为何会迷路、又如何快速重新定向等日常体验，开启了新的研究路径。

中文翻译：

动物如何构建方向感

引言

在印度洋一座偏远岛屿上，六只被密切监测的蝙蝠飞向繁星点点的夜空。当它们飞越这片七英亩的弹丸之地时，植入其大脑的装置将数据实时传回地面一群睡眼惺忪的神经科学家。研究人员试图弄清这些大脑与人类相似的飞行哺乳动物，如何在陌生环境中形成方向感。

发表于《科学》杂志的这项研究指出，蝙蝠利用脑细胞网络来感知岛屿周围的方向。它们的“内在罗盘”既不依赖地磁场，也不参照星空，而是通过识别地标来构建环境心理地图。

这些首次在野外开展的哺乳动物地图构建实验，证实了数十年实验室研究的成果，并为关于神经罗盘如何锚定环境的两大对立理论之一提供了支持。

“我们正在自然真实条件下理解哺乳动物大脑运作的基本原理，”英国斯特灵大学空间导航研究专家、未参与该研究的保罗·杜琴科表示，“这将成为未来五十年人们持续讨论的经典论文。”

尚未发表的后续实验显示，在野外环境中，其他对导航至关重要的细胞比实验室条件下编码更多信息，这凸显了在真实世界检验神经生物学理论的必要性。

神经科学家认为，由“头部方向细胞”构成的类似内在罗盘可能也存在于人类大脑中——尽管尚未被定位。若有朝一日发现它们，这套机制或能解释“转向迷失”后迅速重定向的常见体验，甚至揭示某些人方向感极差的原因。

方向感的奥秘

半个多世纪以来，哺乳动物大脑如何在环境中导航始终令科学家着迷。杜克大学神经科学家南希亚·苏塔纳指出，相关研究已催生“多项极其有趣的现象，其中数项荣获诺贝尔奖”。

上世纪70年代初，伦敦大学学院的约翰·奥基夫在大鼠海马体（大脑记忆中枢）中发现对特定位置产生反应的“位置细胞”。数十年后，挪威科技大学的梅-布里特·莫泽和爱德华·莫泽在相邻脑区发现构建大脑坐标系的“网格细胞”。这三位研究者因相关发现荣获诺贝尔奖。

这两类细胞共同构成动物周围环境的地图。但达特茅斯学院神经科学家杰弗里·陶布指出：“仅知道自身位置不足以实现导航，还需明确朝向方向。二者缺一不可。”

1984年，纽约州立大学布鲁克林分校的吉姆·兰克在研究位置细胞信息离开海马体后的变化时，意外发现了头部方向细胞。陶布（曾师从兰克从事博士后研究）回忆道：“这些细胞不关心动物所处位置，只对动物头部朝向敏感。这是个充满偶然却又无比美妙的发现。”

此后数十年，科学家逐步阐明头部方向细胞在啮齿动物中的工作原理。这些神经元通过视觉、听觉、触觉接收外部信息，同时通过内耳前庭系统感知头部运动。当动物移动时，它会追踪自身相对于地标的运动轨迹，学习将特定地标与方向关联，并据此持续更新心理地图。这套系统现被神经科学家称为头部方向回路或内在罗盘。

“这不是磁力意义上的罗盘，却是绝对意义上的方向指示器，”杜琴科解释道，“罗盘的功能是什么？就是保持相对于当前位置、姿态或环境的方位指向。”

这些头部方向细胞通过环形吸引子网络连接。在哺乳动物中，该网络并非物理环形结构（果蝇中却奇妙地存在实体环形），但可用环形示意图表示。这个环形网络始终处于激活状态：当动物朝向特定方向时，环中特定细胞放电；当动物转向时，这些细胞关闭而其他细胞接续激活。

约翰斯·霍普金斯大学神经科学家詹姆斯·尼里姆（未参与新研究，但为《科学》杂志相关论文撰写评论）说明：“当动物头部旋转360度时，不同细胞会按序激活，每个细胞对应特定方向。”

尼里姆指出，核心问题在于：在动物实际生存的广阔领域里，这些细胞是否会像磁罗盘那样保持固定指向？先前研究衍生出两种对立理论：“全局罗盘假说”主张每个头部方向细胞在穿越大环境时保持固定指向；“镶嵌假说”则认为当动物进入大环境的不同区域时，头部方向细胞会重置方向参数。

此前所有相关研究均在狭小封闭空间进行。要理解罗盘的真实运作机制，科学家必须走向户外。

天然实验室

我们对哺乳动物导航时大脑活动的认知皆来自实验室，但这远非全貌。以色列魏茨曼科学研究所行为系统神经科学家纳胡姆·乌兰诺夫斯基指出，实验室小盒子让动物“瞬间尽览所有景物，这并非挑战性意义上的真实导航，就像在城市中穿行那样”。

城市导航需要我们持续整合时空信息与记忆。我们不仅需要心理地图，还要应对环境干扰：躲避自行车、抢在红灯前过街、绕开垃圾避免撞人。我们需要从未知地点规划路线，还要理解蜿蜒人行道、多径公园、五层公寓等迥异环境如何相互关联。

这种复杂环境难以在实验室模拟，而户外非受控研究更为困难。乌兰诺夫斯基坦言：“尽管神经导航基础研究令人振奋，但位置细胞、网格细胞、头部方向细胞都未在真实户外环境被研究过。我多年梦想实现这个目标，却始终束手无策。”

2016年，他的团队在研究所建造200米长隧道，开发无线系统记录埃及果蝠飞行时的脑活动。《科学》杂志报道显示，位置细胞在隧道中的表现与实验室截然不同——这暗示复杂实验环境才是理解哺乳动物导航的关键。

但隧道对乌兰诺夫斯基仍显局限。2018年在澳大利亚大堡礁潜水时，他灵光乍现：“置身岛屿让我顿悟解决方案——何不在世界某处寻找岛屿作为野外实验室？”

他需要寻找远离大陆（防止蝙蝠逃逸造成生态问题）、大小适中、无人居住、植被稀疏（避免蝙蝠藏身高树）、非自然保护区（规避许可问题）的岛屿。乌兰诺夫斯基坦言：“同时满足这些条件的岛屿极其罕见。”团队筛选全球三四十个候选岛屿，最终锁定坦桑尼亚以东25英里印度洋上的莱瑟姆岛——这是他们研究的埃及果蝠栖息范围内唯一符合条件的岛屿。

这座约四个足球场大小的岛屿，既便于研究人员管控追踪，又确保蝙蝠无法一眼望穿全岛。

乌兰诺夫斯基团队准备观察蝙蝠如何学习在更接近进化环境的复杂栖息地导航。他们在六只埃及果蝠脑中植入数微米细的微丝记录神经活动，数据通过连接记录仪存储。团队乘船携带帐篷、桌椅、发电机、冰箱等数周生活物资登岛，通常在夜间释放蝙蝠并追踪其飞行轨迹。每夜结束时重新捕获蝙蝠下载头部方向细胞及其他导航细胞活动数据。历经2023-2024两个实验季，团队最终获得301次飞行数据。

最初几夜，当蝙蝠开始探索莱瑟姆岛时，其头部方向细胞放电模式粗糙：有些在面向大致南方时激活，有些对应东、西、北方。但到第五六夜，随着对环境日渐熟悉，细胞放电模式趋于稳定，精准对应特定方向且不随岛上位置改变。

由于无法一眼览尽全岛，蝙蝠大脑似乎将岛屿局部信息拼接成完整全局地图。研究结果支持“全局罗盘假说”——这与某些实验预测相符。杜琴科认为这合乎逻辑：“罗盘就该是罗盘。当你进入隔壁房间，它仍应指向正确方向。”

这些细胞如何锚定特定方向？它们并非参照天体线索：当月亮横越天空或被云层遮蔽时，蝙蝠脑活动保持稳定。乌兰诺夫斯基团队的初步实验也排除了地磁场参照的可能。研究团队推测，蝙蝠通过海岸线、实验帐篷、栖息处等地标锚定方向。随着对新空间的熟悉，这些地标融入其内在地图，并引导头部方向细胞激活。

这些发现证实了数十年实验室研究关于头部方向细胞系统在小环境中运作的推测。尼里姆评价：“细胞在广阔自然环境中是否表现相同曾是个悬而未决的问题。这项研究在远超实验模拟规模的复杂野外空间记录细胞活动，在神经科学领域堪称开创。”

超越岛屿的研究

这种真实世界研究已结出硕果。2025年11月圣迭戈神经科学学会会议上，乌兰诺夫斯基展示的早期数据显示：在莱瑟姆岛导航的蝙蝠脑细胞比实验室编码更多信息——例如位置细胞不仅记录位置，还会根据飞行速度激活。

杜琴科认为这些初步发现“为开展自然实验提供了更强论据，提示了神经科学研究的新范式”。他主张神经科学家不应回避复杂性，而应积极拥抱。

随着研究视野超越实验室，科学家也希望突破鼠蝠局限。在城市中穿行的你，必然运用过自身的头部方向系统。尼里姆回忆在曼哈顿行走时自认为向东：“当我拐过街角，期待看到第二大道却出现列克星敦大道时，整个认知世界瞬间翻转——我能真切感受到体内的方向重构。”当意识到内在地图错位时，他能感受到心理空间追赶物理空间时的方向扭转。

人类方向感的神经基础仍知之甚少。头部方向细胞虽未被定位，但存在迹象。杜琴科表示：“我们与啮齿动物和蝙蝠拥有相同脑结构，认为这些结构功能相似并非妄断。”我们在环境导航中的体验无疑表明人类具有方向感（只是程度各异）。

杜克大学神经科学家苏塔纳指出，人类研究的缺失是“我们正努力填补的重大空白”。在癫痫患者知情同意下，她的团队将新设备连接至患者术前监测的植入电极，首次在人类受试者中记录其探索医院房间走廊时导航细胞的活动，收集人体移动时导航细胞追踪身体与头部的数据。

“进入更野性自然的环境，确实能让我们发现实验室永远无法观测的现象，”苏塔纳表示。虽然15分钟的医院走廊探索不算真正的野外，但她的团队正致力于在更复杂环境中记录高分辨率脑活动。“或许不是偏远岛屿，但未来谁说得准呢？”

英文来源：

How Animals Build a Sense of Direction
Introduction
On a remote island in the Indian Ocean, six closely watched bats took to the star-draped skies. As they flew across the seven-acre speck of land, devices implanted in their brains pinged data back to a group of sleepy-eyed neuroscientists monitoring them from below. The researchers were working to understand how these flying mammals, who have brains not unlike our own, develop a sense of direction while navigating a new environment.
The research, published in Science, reported that the bats used a network of brain cells that informed their sense of direction around the island. Their “internal compass” was tuned by neither the Earth’s magnetic field nor the stars in the sky, but rather by landmarks that informed a mental map of the animal’s environment.
These first-ever wild experiments in mammalian mapmaking confirm decades of lab results and support one of two competing theories about how an internal neural compass anchors itself to the environment.
“Now we’re understanding a basic principle about how the mammalian brain works” under natural, real-world conditions, said the behavioral neuroscientist Paul Dudchenko, who studies spatial navigation at the University of Stirling in the United Kingdom and was not involved in the study. “It will be a paper people will be talking about for 50 years.”
Follow-up experiments that haven’t yet been published show that other cells critical to navigation encode much more information in the wild than they do in the lab, emphasizing the need to test neurobiological theories in the real world.
Neuroscientists believe that a similar internal compass, composed of neurons known as “head direction cells,” might also exist in the human brain — though they haven’t yet been located. If they are someday found, the mechanism could shed light on common sensations such as getting “turned around” and quickly reorienting oneself. It might even explain why some of us are so bad at finding our way.
A Sense of Direction
How the mammalian brain navigates the environment has been a source of fascination for scientists for at least half a century. Its study has led to the discovery of “extremely interesting phenomena, several of which have won Nobel Prizes,” said Nanthia Suthana, a neuroscientist at Duke University.
In the early 1970s, John O’Keefe, a neuroscientist at University College London, discovered cells in the rat hippocampus, the brain’s memory hub, that responded to specific locations in the rodents’ enclosures. He called them “place cells.” A few decades later, May-Britt Moser and Edvard Moser of the Norwegian University of Science and Technology discovered, in a nearby brain area, cells that create a coordinate system for the brain, which they called “grid cells.” The three researchers were awarded a Nobel Prize for their discoveries.
Together, these two cell types can create a map of an animal’s surroundings. But knowing where you are in space isn’t enough to get you somewhere else. “You also need to know what direction you’re facing,” said Jeffrey Taube, a neuroscientist at Dartmouth College. “You need those two key pieces of information. One without the other doesn’t do you much good.”
In 1984, Jim Ranck, a neuroscientist at the State University of New York Downstate in Brooklyn, New York, was investigating what happens when information from place cells leaves the hippocampus when he accidentally discovered what became known as head direction cells. These cells didn’t seem to care where the animal was located; instead, they responded to the direction the animal was facing. “It was a very serendipitous but obviously wonderful finding,” said Taube, who did his postdoctoral work under Ranck.
In the years since, neuroscientists have characterized how head direction cells work in rodents. The neurons receive inputs from the external world, through the things we see, hear, and touch, and also from the internal world, especially from the vestibular system, a network in the inner ear that tracks head movements. It’s thought that as an animal moves around, it keeps track of its movement relative to the landmarks around it, learns to associate certain landmarks with certain directions, and uses this information to constantly update its mental map. Neuroscientists have come to call this system the head direction circuit, or internal compass.
“It’s not a compass in a magnetic sense, but it is a compass in an absolute sense,” Dudchenko said. “What does a compass do? It keeps orientation relative to where you are, or where you’re standing, or what environment you are in.”
These head direction cells are connected in a ringlike system called a ring attractor network. In mammals, this network is not a physical ring (though it is, strangely, in fruit flies), but it can be schematically represented as such. The ring is always active. When an animal faces a particular direction, certain cells in the ring fire. When the animal turns, those cells turn off and others activate in a continuous fashion.
“As the animal keeps turning its head 360 degrees, a sequence of different cells will fire, each of them tuned to a specific direction,” said James Knierim, a neuroscientist at Johns Hopkins University who was not involved in the new research. (He co-authored an accompanying perspective on the paper for Science.)
The big question, Knierim said, was whether these cells would remain faithful to their assigned directions, as a magnetic compass does, in the real world, where animals live in large territories. Previous work had generated two competing theories. The “global compass” hypothesis claims that each head direction cell commits to a direction during continuous navigation through a large environment: A cell that fires when an animal faces northeast will always fire for northeast. The “mosaic” hypothesis suggests that head direction cells reset and change their compass direction as an animal moves through different regions of a large environment, so that north-indicating cells in one region may represent east in another part.
All the research on this question had been done in small, enclosed spaces. To understand how the compass really works, the scientists needed to go outside.
A Natural Laboratory
Everything we know about what’s going on in the brains of mammals as they navigate their environments comes from lab experiments. But they give an incomplete view. In a small box on a lab bench, an animal sees “immediately everything there is to see,” said Nachum Ulanovsky, a behavioral systems neuroscientist at the Weizmann Institute of Science in Israel. “It’s not real navigation in the challenging sense, like you would navigate in a city.”
Courtesy of Nachum Ulanovsky
When walking around a city, on the other hand, we constantly integrate information about space and time, and from our own memories. We need a mental map, sure, but we also must deal with environmental interference: We need to avoid a cyclist, run across a street before the light turns red, and step over trash without slamming into other people. We need to know how to get from point A to point B, even if we’ve never been there before. And we need to know how vastly different environments — meandering sidewalks, a park with many trails, a fifth-floor apartment — connect to one another.
This kind of complex environment is hard to simulate in the lab. But studying the sense of direction outside the lab, in an uncontrolled setting, can be even harder. So, despite the excitement around the neural basis of navigation, “none of these neurons — neither place cells, nor grid cells, nor head direction cells — had been studied in the real world, outdoors,” Ulanovsky said. “So I had, for many years, this dream that we would like to do that. But for years, it stayed as a dream because how do you even approach this?”
In 2016, his team built a 200-meter-long tunnel at the Weizmann Institute and developed wireless systems to record the brain activity of Egyptian fruit bats as they flew through it. The team reported in Science that place cells behaved differently in the tunnels than they had in the lab — a hint that a more complex experimental environment would be key to really understanding mammalian navigation.
Nachum Ulanovsky
But a tunnel was still too confined for Ulanovsky. He wanted to create conditions closer to the real world. The answer came to him in 2018 as he was scuba diving on the Great Barrier Reef in Australia. “Being on an island there, it hit me that that’s a solution,” he recalled. “Suppose I find an island somewhere in the world” to use as a wild laboratory.
He searched for an island far away from land (so his bats couldn’t escape and create ecological problems) that was not too big and not too small. It had to be uninhabited by people and mostly barren (so bats wouldn’t hide in tall trees), and it couldn’t be a nature reserve (to avoid permitting issues). “The conjunction of these things is pretty rare,” Ulanovsky said. His team homed in on 30 or 40 islands across the world that might work. Only one was in the home range of the Egyptian fruit bats they study: Latham Island, in the Indian Ocean 25 miles east of Tanzania.
Latham Island, a plot of land the size of about four soccer fields, was small enough for the researchers to contain and track the bats — and big enough to ensure that the bats couldn’t see from one end to the other.
Ulanovsky’s team was ready to watch as bats learned to navigate a complex habitat more like the one they evolved in. They implanted microwires, each a few micrometers thick, in the brains of six Egyptian fruit bats to record neural activity; the wires connected to a data logger, which stored the data. They brought the bats to the island on a boat, along with everything the scientists needed to sustain themselves for a few weeks, including tents, chairs, tables, generators, and refrigerators. They released the bats, usually at night, and tracked their positions as they flew across the island. At the end of every night, the researchers re-captured the bats to download data on the activity of head direction cells and other cells involved in navigation. By the end of the experiments, performed over two seasons in 2023 and 2024, the researchers had data from 301 flights.
Palgi and Orian Las
On the first couple of nights, as the bats began to explore Latham Island, their head direction cells fired crudely. Some fired when the bats faced generally south, others while they faced generally east, west or north. But by night five or six, as the place grew more familiar to them, the cells had stabilized to fire in coordination with precise directions and did not change depending on where the animal was on the island.
Because they could not see the entire island at once, their brains seemed to be stitching together small parts of the island into a global whole. The findings suggest that the global compass hypothesis is indeed correct, as some experiments have predicted. This makes sense, as “a compass should be a compass,” Dudchenko said. “If you move to the next room, it should still be pointing in the right direction.”
How did these cells anchor themselves to particular directions? They weren’t adjusting to celestial cues; the bats’ brain activity remained stable as the moon moved across the sky and when the moon and stars were covered by clouds. Nor were the head direction cells anchoring themselves to the Earth’s magnetic field, as some preliminary experiments by Ulanovsky’s team had suggested. The team hypothesizes that the bats anchored themselves to landmarks in their environment, such as the coastline, the experimenters’ tents, and their perches. As they got to know the new space, the landmarks became part of their internal maps and cued the head direction cells to fire.
The findings confirmed decades of lab work suggesting how this head direction cell system worked in smaller environments. “It was an open question, one way or the other, whether the cells behaved the same way in large, natural environments,” Knierim said. He and others applauded the study for recording the activity of these cells out in the wild, in a much bigger and more complex space than experiments could simulate. “In this area of neuroscience, there’s just nothing like that,” he said.
Beyond the Island
Already, this real-world approach is bearing fruit. In November 2025, at the Society for Neuroscience meeting in San Diego, Ulanovsky presented early data showing that the brain cells of bats navigating Latham Island encoded more information than they do inside the lab — for example, place cells not only recorded the bat’s location but also activated based on how fast the bat was going.
Yuval Barkai
These preliminary findings make an “even better argument for doing natural experiments,” Dudchenko said. “They suggest a new approach to how we do neuroscience.” Instead of crafting experiments that control for complexity, neuroscientists should embrace it, he said.
As neuroscientists look beyond the lab, they’re also hoping to look beyond rats and bats. If you’ve spent any time navigating a city, you’ve surely employed your own head direction system. Knierim recalls walking in Manhattan; he thought he was heading east. “When I hit the corner, and I’m expecting to see Second Avenue, and I see Lexington Avenue [instead] — my whole head, you know, my own perception of the world just spun around,” he said. “I can literally feel it inside.” When he realized his internal map was misaligned, he could feel it twist around him as his mental space caught up with his physical one.
Not much is known about the neural basis of our own sense of direction. Head direction cells have not yet been located in humans, though there is some evidence that they exist. “We do have the same brain structure [as rodents and bats], so it’s not too crazy to think that those brain structures then have similar function,” Dudchenko said. Certainly, our experiences navigating our environments suggest that we have a sense of direction (some more than others).
The lack of human studies is a “major gap that we’re trying to fill,” said Suthana, the Duke neuroscientist. With consent from epilepsy patients, Suthana and her team connected a new device to electrodes already implanted into their brains for presurgical monitoring. Then she recorded navigation cells in humans exploring a seminatural environment — a hospital room and hallway — to collect data on how navigational cells track the body and head as a person moves. This was the first time such a study had been performed in human subjects.
“Moving into these wilder, naturalistic environments really has the ability for us to test things or find things we would never see in the lab,” she said. While 15 minutes wandering a hospital hallway isn’t exactly the wild, her team is working toward the goal of recording high-resolution brain activity in even more complex environments. “Maybe not on a remote island, but who knows?”